Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Transfer Learning Model”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 271 records · Page 15
All sourcesNASA NTRSDOE OSTI
DOE OSTI2026

What do physics-informed DeepONets learn? Understanding and improving training for scientific computing applications

Physics-informed deep operator networks (DeepONets) have emerged as a promising approach toward numerically approximating the solution of partial differential equations (PDEs). In this work, we aim to develop further understanding of what is being learned by physics-informed DeepONets by assessing the universality of the extracted basis functions and demonstrating their potential toward model reduction with spectral methods. Results provide clarity about measuring the performance of a physics-informed DeepONet through the decays of singular values and expansion coefficients. In addition, we propose a transfer learning approach for improving training for physics-informed DeepONets between parameters of the same PDE as well as across different, but related, PDEs where these models struggle to train well. This approach results in significant error reduction and learned basis functions that are more effective in representing the solution of a PDE.

Deep operator networks↗
DOE OSTI2022

Classification of animal sounds in a hyperdiverse rainforest using convolutional neural networks with data augmentation

To protect tropical forest biodiversity, we need to be able to detect it reliably, cheaply, and at scale. Automated detection of sound producing animals from passively recorded soundscapes via machine-learning approaches is a promising technique towards this goal, but it is constrained by the necessity of large training data sets. Using soundscapes from a tropical forest in Borneo and a Convolutional Neural Network model (CNN), we investigate i) the minimum viable training data set size for accurate prediction of call types (‘sonotypes’), and ii) the extent to which data augmentation and transfer learning can overcome the issue of small and imbalanced training data sets. We found that even relatively high sample sizes (>80 per sonotype) lead to mediocre accuracy, which however improved significantly with data augmentation and transfer learning, including at extremely small sample sizes (3 per sonotype), regardless of taxonomic group or call characteristics. Neither transfer learning nor data augmentation alone achieved high accuracy. Our results suggest that transfer learning and data augmentation could make the use of CNNs to classify species’ vocalizations feasible even for small soundscape-based projects with many rare species. Retraining our open-source model requires only basic programming skills which makes it possible for individual conservation initiatives to match their local context, in order to enable more evidence-informed management of biodiversity.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Fouling modeling and prediction approach for heat exchangers using deep learning

In this article, we develop a generalized and scalable statistical model for accurate prediction of fouling resistance using commonly measured parameters of industrial heat exchangers. This prediction model is based on deep learning where a scalable algorithmic architecture learns non-linear functional relationships between a set of target and predictor variables from large number of training samples. Here, the efficacy of this modeling approach is demonstrated for predicting fouling in an analytically modeled cross-flow heat exchanger, designed for waste heat recovery from flue-gas using room temperature water. The performance results of the trained models demonstrate that the mean absolute prediction errors are under 10 –4 KW –1 for flue-gas side, water side and overall fouling resistances. The coefficients of determination (R 2 ), which characterize the goodness of fit between the predictions and observed data, are over 99%. Even under varying levels of measurement noise in the inputs, we demonstrate that predictions over an ensemble of multiple neural networks achieves better accuracy and robustness to noise. We find that the proposed deep-learning fouling prediction framework learns to follow heat exchanger flow and heat transfer physics, which we confirm using locally interpretable model agnostic explanations around randomly selected operating points. Overall, we provide a robust algorithmic framework for fouling prediction that can be generalized and scaled to various types of industrial heat exchangers.

42 ENGINEERING↗
DOE OSTI2022

Explosion Detection using Transfer Learning via YAMNet [Poster]

The acoustic data of explosions and noise collected on smartphones were fed to the YAMNet model to obtain the scores of each class. The data was split 60/20/20 for training, validation, and testing. A single-layered neural network was trained using the computed scores to predict if the data was an explosion or noise. The trained model was then combined with the YAMNet model. The combined model was tested with the ESC-50 dataset to investigate overall accuracy and class based false positive rates.

98 - NUCLEAR DISARMAMENT, SAFEGUARDS, AND PHYSICAL↗
DOE OSTI2023

An interlaboratory comparison of mid-infrared spectra acquisition: Instruments and procedures matter

Diffuse reflectance spectroscopy has been extensively employed to deliver timely and cost-effective predictions of a number of soil properties. However, although several soil spectral laboratories have been established worldwide, the distinct characteristics of instruments and operations still hamper further integration and interoperability across mid-infrared (MIR) soil spectral libraries. In this study, we conducted a large-scale ring trial experiment to understand the lab-to-lab variability of multiple MIR instruments. By developing a systematic evaluation of different mathematical treatments with modeling algorithms, including regular preprocessing and spectral standardization, we quantified and evaluated instruments' dissimilarity and how this impacts internal and shared model performance. We found that all instruments delivered good predictions when calibrated internally using the same instruments' characteristics and standard operating procedures by solely relying on regular spectral preprocessing that accounts for light scattering and multiplicative/additive effects, e.g., using standard normal variate (SNV). When performing model transfer from a large public library (the USDA NSSCKSSL MIR library) to secondary instruments, good performance was also achieved by regular preprocessing (e. g., SNV) if both instruments shared the same manufacturer. However, significant differences between the KSSL MIR library and contrasting ring trial instruments responses were evident and confirmed by a semi-unsupervised spectral clustering. For heavily contrasting setups, spectral standardization was necessary before transferring prediction models. Non-linear model types like Cubist and memory-based learning delivered more precise estimates because they seemed to be less sensitive to spectral variations than global partial least square regression. In summary, the results from this study can assist new laboratories in building spectroscopy capacity utilizing existing MIR spectral libraries and support the recent global efforts to make soil spectroscopy universally accessible with centralized or shared operating procedures.

58 GEOSCIENCES↗
DOE OSTI2024

Polyalkenamers as Drop-In Additives for Ring-Opening Metathesis Polymerization: A Promising Upcycling Paradigm

Here we report a distinct strategy to upcycle waste polyalkenamers such as polybutadiene into new, performance-advantaged materials by using them as drop-in additives for ring-opening metathesis polymerization (ROMP). The polyalkenamers serve as competent chain-transfer agents in ROMPs of common classes of cyclic olefin monomers, facilitating good molecular weight control, allowing low Ru catalyst loadings, and enabling efficient incorporation of the polyalkenamer into the synthesized polymeric material. We successfully demonstrate ROMP using model polyalkenamers and translate these learnings to leverage commercial polybutadiene and acrylonitrile butadiene styrene (ABS) as chain transfer agents for ROMP copolymerizations. Critically, our strategy is shown to be highly efficient and operationally simple, quantitatively incorporating the polyalkenamer and inheriting aspects of its thermomechanical performance. Our results highlight a promising pathway for the upcycling of polyalkenamers and provide an alternative to existing deconstruction and functional upcycling strategies.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2022

Using soil library hyperspectral reflectance and machine learning to predict soil organic carbon: Assessing potential of airborne and spaceborne optical soil sensing

Soil organic carbon (SOC) is a key variable to determine soil functioning, ecosystem services, and global carbon cycles. Spectroscopy, particularly optical hyperspectral reflectance coupled with machine learning, can provide rapid, efficient, and cost-effective quantification of SOC. However, how to exploit soil hyperspectral reflectance to predict SOC concentration, and the potential performance of airborne and satellite data for predicting surface SOC at large scales remain relatively underknown. Here, this study utilized a continental-scale soil laboratory spectral library (37,540 full-pedon 350–2500 nm reflectance spectra with SOC concentration of 0–780 g·kg –1 across the US) to thoroughly evaluate seven machine learning algorithms including Partial-Least Squares Regression (PLSR), Random Forest (RF), K-Nearest Neighbors (KNN), Ridge, Artificial Neural Networks (ANN), Convolutional Neural Networks (CNN), and Long Short-Term Memory (LSTM) along with four preprocessed spectra, i.e. original, vector normalization, continuum removal, and first-order derivative, to quantify SOC concentration. Furthermore, by using the coupled soil-vegetation-atmosphere radiative transfer model, we simulated twelve airborne and spaceborne hyper/multi-spectral remote sensing data from surface bare soil laboratory spectra to evaluate their potential for estimating SOC concentration of surface bare soils. Results show that LSTM achieved best predictive performance of quantifying SOC concentration for the whole data sets (R 2 = 0.96, RMSE = 30.81 g·kg –1 ), mineral soils (SOC ≤ 120 g·kg –1 , R 2 = 0.71, RMSE = 10.60 g·kg –1 ), and organic soils (SOC > 120 g·kg –1 , R 2 = 0.78, RMSE = 62.31 g·kg –1 ). Spectral data preprocessing, particularly the first-order derivative, improved the performance of PLSR, RF, Ridge, KNN, and ANN, but not LSTM or CNN. We found that the SOC models of mineral and organic soils should be distinguished given their distinct spectral signatures. Finally, we identified that the shortwave infrared is vital for airborne and spaceborne hyperspectral sensors to monitor surface SOC. This study highlights the high accuracy of LSTM with hyperspectral/multispectral data to mitigate a certain level of noise (soil moisture <0.4 m 3 ·m –3 , green leaf area < 0.3 m 2 ·m –2 , plant residue <0.4 m 2 ·m –2 ) for quantifying surface SOC concentration. Forthcoming satellite hyperspectral missions like Surface Biology and Geology (SBG) have a high potential for future global soil carbon monitoring, while high-resolution satellite multispectral fusion data can be an alternative.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Supervised enhancer prediction with epigenetic pattern recognition and targeted validation

Enhancers are important non-coding elements, but they have traditionally been hard to characterize experimentally. The development of massively parallel assays allows the characterization of large numbers of enhancers for the first time. Here, we developed a framework using Drosophila STARR-seq to create shape-matching filters based on meta-profiles of epigenetic features. We integrated these features with supervised machine-learning algorithms to predict enhancers. We further demonstrated that our model could be transferred to predict enhancers in mammals. We comprehensively validated the predictions using a combination of in vivo and in vitro approaches, involving transgenic assays in mice and transduction-based reporter assays in human cell lines (153 enhancers in total). The results confirmed that our model can accurately predict enhancers in different species without re-parameterization. Finally, we examined the transcription factor binding patterns at predicted enhancers versus promoters. Here, we demonstrated that these patterns enable the construction of a secondary model that effectively distinguishes enhancers and promoters.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2024

Identifying atmospheric rivers and their poleward latent heat transport with generalizable neural networks: ARCNNv1

Abstract. Atmospheric rivers (ARs) are extreme weather events that can alleviate drought or cause billions of US dollars in flood damage. By transporting significant amounts of latent energy towards the poles, they are crucial to maintaining the climate system's energy balance. Since there is no first-principle definition of an AR grounded in geophysical fluid mechanics, AR identification is currently performed by a multitude of expert-defined, threshold-based algorithms. The variety of AR detection algorithms has introduced uncertainty into the study of ARs, and the thresholds of the algorithms may not generalize to new climate datasets and resolutions. We train convolutional neural networks (CNNs) to detect ARs while representing this uncertainty; we name these models ARCNNs. To detect ARs without requiring new labeled data and labor-intensive AR detection campaigns, we present a semi-supervised learning framework based on image style transfer. This framework generalizes ARCNNs across climate datasets and input fields. Using idealized and realistic numerical models, together with observations, we assess the performance of the ARCNNs. We test the ARCNNs in an idealized simulation of a shallow-water fluid in which nearly all the tracer transport can be attributed to AR-like filamentary structures. In reanalysis and a high-resolution climate model, we use ARCNNs to calculate the contribution of ARs to meridional latent heat transport, and we demonstrate that this quantity varies considerably due to AR detection uncertainty.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2026

Maximizing machine learning interatomic potential transferability for the discovery of the novel stellated octadecagon Bi18-Pt24 cage structure

Achieving true transferability remains the central challenge for Machine Learning Interatomic Potentials (ML-IAPs) in modeling complex bimetallic nanoclusters across their vast potential energy surfaces. We systematically investigate data selection strategies to optimize the Chebyshev Interaction Model for Efficient Simulation (ChIMES) potential for the Bi-Pt nanoclusters by comparing three innovative sampling methods: Principal Component Analysis (PCA)/k-means (structural diversity), t-distributedStochasticNeighborEmbedding (t-SNE)/k-means (force-space diversity), and hierarchical clustering. Quantitatively, the PCA/k-means strategy proved most effective for global accuracy, yielding the lowest force errors and achieving energy root mean square errors (RMSE) values competitive with Density Functional Theory (DFT), demonstrating excellent accuracy (19.16meV/atom). Structural validation on 34 unique DFT-optimized isomers further confirmed the potential’s high fidelity, with the best model PCA/k-means reproducing structures with an average root mean square deviation (RMSD) of 0.10 Å. However, the t-SNE methods, by maximizing diversity in the force space, demonstrated superior extrapolative power, leading to the more precise prediction of a novel stellated octadecagon Bi18⁢Pt24 cage structure, demonstrating the potential for exploring previously unseen morphologies. Our results establish a clear methodology for strategic data sampling that successfully maximizes ML-IAP transferability, providing an accurate and computationally efficient tool that accelerates the theoretical discovery of complex bimetallic architectures.

Vangheluwe, Raphaël [Université Paris-Saclay, CNRS↗
DOE OSTI2026

Evaluating the Limits of QAOA Parameter Transfer at High-Rounds on Sparse Ising Models With Geometrically Local Cubic Terms

The emergent practical applicability of the Quantum Approximate Optimization Algorithm (QAOA) for approximate combinatorial optimization is a subject of considerable interest. One of the primary limitations of QAOA is the task of finding a set of good parameters, which is usually done using a variational optimization loop. Parameter transfer, or parameter concentration, is a phenomenon where QAOA angles trained on problem instances that are self-similar tend to perform well for other problem instances from that similar class. This suggests a potentially highly efficient and scalable non-variational learning method for QAOA angle finding. In this work, we systematically study QAOA parameter transferability from small problem sizes (16 and 27 decision variables) onto large problem instances (up to 156 qubits) for heavy-hex graph Ising models with geometrically local higher order terms using the Julia based QAOA simulation tool \texttt{JuliQAOA} to perform classical angle finding for up to $49$ QAOA layers ($p$). Parameter transfer of the fixed angles is validated using a combination of full statevector, Projected Entangled Pair States (PEPS), Matrix Product State (MPS), and LOWESA numerical simulations. We find that the QAOA parameter transfer from single instances applied to other (unseen) problem instances does not in general provide monotonically improving performance as a function of $p$ - there are many cases where the performance temporarily decreases as a function of $p$ - but despite this the transferred angles have a general trend of improved expectation value as the QAOA depth increases, in many cases converging close to the true ground-state energy of the $100+$ qubit instances. We also sample the hardware-compatible Ising models using the ensemble of transfer-learned QAOA parameters on several superconducting qubit IBM Quantum processors with 127, 133, and 156 qubits. We find continuous solution quality improvement of the hardware-compatible QAOA circuits run on the IBM NISQ processors up to $p=5$ on \texttt{ibm\_fez}, up to $p=9$ on \texttt{ibm\_torino}, and up to $p=10$ on \texttt{ibm\_pittsburgh}.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2024

Machine learning analysis of high-repetition-rate two-dimensional Thomson scattering spectra from laser-produced plasmas

With the emergence of high-repetition-rate two-dimensional Thomson scattering (TS) measurements, improving spectral data analysis is a key area of interest. Here, we present a new way to derive the electron temperature and density of laser-driven blast waves in plasmas from their TS spectra with machine learning (ML). This analysis occurs in both the non-collective (α < 1) and collective (α > 1) scattering regimes with the goal of autonomously and more accurately determining T c and n e both where spectral data has been collected and to give the ability to predict these attributes in regions where data has not been collected. We introduce three ML models, one trained only on experimental data, one only on synthetic data, and one using transfer learning, and compare their speed and accuracy with the conventional TS inversion algorithms in the open source PlasmaPy python package.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2025

TPCpp-10M: Simulated proton-proton collisions in a time projection chamber for AI foundation models

Scientific foundation models hold great promise for advancing nuclear and particle physics by improving analysis precision and accelerating discovery. Yet, progress in this field is often limited by the lack of openly available large scale datasets, as well as standardized evaluation tasks and metrics. Furthermore, the specialized knowledge and software typically required to process particle physics data pose significant barriers to interdisciplinary collaboration with the broader machine learning community. This work introduces a large, openly accessible dataset of 10 million simulated proton-proton collisions, designed to support self-supervised training of foundation models. To facilitate ease of use, the dataset is provided in a common NumPy format. In addition, it includes 70,000 labeled examples spanning three well defined downstream tasks: track finding, particle identification, and noise tagging, to enable systematic evaluation of the foundation model's adaptability. The simulated data are generated using the Pythia Monte Carlo event generator at a center of mass energy of $\sqrt{s}$ = 200 GeV and processed with Geant4 to include realistic detector conditions and signal emulation in the sPHENIX Time Projection Chamber at the Relativistic Heavy Ion Collider, located at Brookhaven National Laboratory. This dataset resource establishes a common ground for interdisciplinary research, enabling machine learning scientists and physicists alike to explore scaling behaviors, assess transferability, and accelerate progress toward foundation models in nuclear and high energy physics. The complete simulation and reconstruction chain is reproducible with the sPHENIX software stack. All data and code locations are provided under Data Accessibility.

Data Analysis, Statistics and Probability (physics↗
DOE OSTI2021

Progressive transfer learning for low-frequency data prediction in full waveform inversion

To effectively overcome the cycle-skipping issue in full waveform inversion (FWI), we developed a deep neural network (DNN) approach to predict the absent low-frequency components by exploiting the hidden physical relation connecting the low- and the high-frequency data. To efficiently solve this challenging nonlinear regression problem, two novel strategies were proposed to design the DNN architecture and to optimize the learning process: (1) dual data feed structure; (2) progressive transfer learning. With the dual data feed structure, not only the high-frequency data, but also the corresponding beat tone data are fed into the DNN to relieve the burden of feature extraction. The second strategy, progressive transfer learning, enables us to train the DNN using a single evolving training dataset. Within the framework of the progressive transfer learning, the training dataset continuously evolves in an iterative manner by gradually retrieving the subsurface information through the physics-based inversion module, progressively enhancing the prediction accuracy of the DNN and propelling the inversion process out of the local minima. Here, the synthetic numerical experiments suggest that, without any a priori geological information, the low-frequency data predicted by the progressive transfer learning are sufficiently accurate for an FWI engine to produce reliable subsurface velocity models free of cycle-skipping artifacts.

02 PETROLEUM↗
DOE OSTI2023

Synergy of semiempirical models and machine learning in computational chemistry

Catalyzed by enormous success in the industrial sector, many research programs have been exploring data-driven, machine learning approaches. Performance can be poor when the model is extrapolated to new regions of chemical space, e.g., new bonding types, new many-body interactions. Another important limitation is the spatial locality assumption in model architecture, and this limitation cannot be overcome with larger or more diverse datasets. The outlined challenges are primarily associated with the lack of electronic structure information in surrogate models such as interatomic potentials. Given the fast development of machine learning and computational chemistry methods, we expect some limitations of surrogate models to be addressed in the near future; nevertheless spatial locality assumption will likely remain a limiting factor for their transferability. Here, we suggest focusing on an equally important effort—design of physics-informed models that leverage the domain knowledge and employ machine learning only as a corrective tool. In the context of material science, we will focus on semi-empirical quantum mechanics, using machine learning to predict corrections to the reduced-order Hamiltonian model parameters. The resulting models are broadly applicable, retain the speed of semiempirical chemistry, and frequently achieve accuracy on par with much more expensive ab initio calculations. These early results indicate that future work, in which machine learning and quantum chemistry methods are developed jointly, may provide the best of all worlds for chemistry applications that demand both high accuracy and high numerical efficiency.

36 MATERIALS SCIENCE↗
DOE OSTI2025

Virtual to Physical: Reinforcement Learning to Optimize SNS Particle Accelerator Controls

Complex accelerators must have control systems that can handle dynamic nonlinear environments. This makes traditional control methods unsuitable as they can struggle to adapt to these uncertainties. This provides an ideal environment for reinforcement learning algorithms as they are adaptable and generalizable. We present a reinforcement learning pipeline that can effectively handle the dynamics of a complex accelerator. We test and prove our pipelines capabilities on multiple environments including the Spallation Neutron Source (SNS) and the Beam Test Facility (BTF) at Oakridge National Lab (ORNL). Due to the limited time available to train an online algorithm like reinforcement learning on a real accelerator, we utilize a virtual twin accelerator (VIRAC) developed by ORNL to pretrain the policy and show its ability to converge in the virtual environment. We then test the adaptability of the pretrained RL model by applying it on the real accelerator and comparing the results. Utilizing our Scientific Optimization and Controls Toolkit (SOCT) and open-source standards such as Gymnasium we create and solve for a MEBT orbit correction problem in the SNS and an emittance maximization problem in the BTF. We show how Twin Delayed Deep Deterministic Policy Gradient (TD3) can solve this optimization environment in the virtual accelerator and transfer this policy onto the real accelerator for inference and model retraining. We show how reinforcement learning can be utilized as a control system for complex accelerators and provide a model pipeline for how an implementation performs and can be adapted to new accelerator control problems.

Kasparian, Armen [Thomas Jefferson National Accele↗